Forgot your password?
typodupeerror

Comment Re:It's not AI's fault, it's bad engineering (Score 1) 88

Exactly, and now everything is so broken, you have to fix it, but don't know how, and can't do it properly. This is why constant upgrades and fixes are required, and that can mean porting code. When my paternal grandfather was in his late 70s, maybe early 80s, the US government called him to fix an old DEC mainframe. That really underscores the level of stupidity and recklessness companies have fallen into.

Comment It's not AI's fault, it's bad engineering (Score 4, Interesting) 88

AI might have found thousands of possible attack vectors, but the patches don't need to take anything offline. Engineering is understanding how things and systems work, operate, and using skill, and knowledge to maintain or better those things / systems. Engineering is not, “AI told us we have an issue, let's push wildly untested or under tested code / fixes into a system we don't understand, and can't manage.”.

The old saying: “If it's not broke, don't fix it.”, is showing why it was always bad advice! It's not broke now, but in 5, 10, 15+ years, will it be broke, and do we need better management and oversight of the system? Think about your favourite framework or language, if it's EOL, do you keep using it? When does an EOL status concern you? Angular, for instance, is in v22, if you're running v18, are you going to push so you're on v20, or v21? If you're on v15, would update?

You're seeing the excuses pile up, X was decades old, not designed for our modern systems, and workflows, and now those AI patches which weren't reviewed, are the bull in a china shop. It wasn't broke, no one took the time to understand, improve, and polish the system, and now it's not only broken, it's stuck in a state it's constantly failing and being band aided together with duct tape, and prayer.

Comment Re: Identify the customers and issue refunds! (Score 1) 99

They illegally collected tariffs that the customers paid, they owe that money back to the customers. If they had absorbed the tariff costs they could have prevented this mess, but they didn't, and so it's on them to fix it. I don't give a rats ass if it costs company Z $10 billion dollars, don't use the tariff refunds, those are reserved for the people who paid the costs.

For decades we've been forced to tolerate grotesque violations of our digital liberties, privacy, and security, in the name of service improvements, wouldn't one of those improvement involve compensating the customers they committed fraud against? How do they do it? I don't care, it's not something I have to consider, and I'm not a US citizen, so it never affected me. The companies caused the problem, and most of them have the means to fix it, if it was considerable enough to warrant fixing, so get to it.

As for the small family shop that doesn't have this information, I don't know, I really don't know, it would likely be near impossible. If they have good record keeping and auditing, then correlate the purchase history and figure it out.

Comment Re:Using AI to help is not it doing all the work (Score 1) 103

Okay, but I still see nothing wrong with using AI as an exploration tool. The resource consumption is irresponsible, copyright is being thoroughly ignored and violated, and it's allowing people to create works far beyond their skill level. That being said, it's great to bounce ideas off, or do feedback document checking, or reading large amount of documentation to find an answer.

Comment Caught my health insurance using AI. (Score 1) 54

I'm in Ontario, Canada.

I'm on cannabinoid-therapy for a chronic pain and a seizure condition, which means I use medical cannabis. The insurance company already clear that, and I get ~$300 / month in cannabis related expensive returned at 80%, which equates to a ~$250 reimbursement. That's been signed off, and I have the forms to prove it. I have 18 doctors treating me, that's not a typo, 18 doctors. I have to travel in February for my wedding out of country, and so my doctors prescribed Sativex, which in Canada has DIN (Drug Identification Number), and goes through a different channel to be approved by insurance.

The first, and second time I submitted my prior approval forms, denied with the reasoning, and I'm not making this up, these are direct quotes:
1. The Canadian medical profession doesn't broadly accept it.
2. Health Canada doesn't approve of it for this type of treatment.
3. This drug doesn’t meet your plan’s criteria for approval (prior authorization).
4. You may want to talk with your physician for alternate treatment options.

Cannabis isn't accepted broadly by the Canadian medical community? It's widely accepted, and considered effective for certain narrow conditions. Most doctors won't prescribe it, but I have rarely met a doctor who won't accept it as a possible solution. Health Canada doesn't approve it? They do, with confusing and circular language, but, they do accept its medical benefits, again, for narrow conditions. The drug must meet the plans criteria, I'm already using a cannabinoid medication, cannabis, which they signed off on. The last point is the most annoying, Sativex is the only medication that is a natural cannabinoid medication, which also has a DIN assigned, specifically for Chronic Neuropathic Pain Disorders. That was followed up with statements that a consultant medical professional had looked up all my information, paper work, medical history and conditions, and determined Sativex was not suitable, and other medications existed.

When I pushed back slightly, the following happened:
1. We can't give you the contact information for the consultant.
2. They used all the available medical information provided.
3. Sativex has not been approved for your conditions (it literally has according to the HC monograph).
4. Cannabinoid-therapy would not consider effective in this circumstance (I'm already using it and approved).

That caused several rounds of phone calls and emails which went nowhere, so I email the CEO, COO, CRO, CPO, CHRO, EVPs, SVPs, JVPs, and head legal counsel for the company. Submitted the same 40-page form submission, which had extensive medical history, and guess what? It got approved, which has the only possible conclusion that the first and second submission were never read by a human.

How do they possibly claim that cannabis wouldn't be effective for chronic pain and seizures? The two most known, accepted, and widely agreed upon uses in a medical setting are seizures and pain. Pair that with the fact I'm already using a cannabinoid-therapy, and Sativex is literally THCd9 and CBD in a formulation accepted by HC and issued a DIN number. Taking all of that into account, and seeing my previous approvals, no human looked at anything in that approval discourse. What did happen? AI read the application, then looked up the old outdated monograph, and concluded that I don't fit the narrow, one off exception, for Sativex approval, dropped years ago. It used to only be approved for MS related chronic neuropathic pain, which HC had to drop after failing to define how MS chronic neuropathic pain was different from any other undifferentiated form of chronic neuropathic pain. The monograph now states it can treat or help with chronic neuropathic pain, but you'd only know that if you bothered to read the monograph listing, which the AI didn't do, and no human intervined with, until I cause a fuss.

Comment Using AI to help is not it doing all the work (Score 1) 103

If you enter a plot, theme, characters, objectives, and other such information into a LLM, and it spits out a book, and you publish it, that's bad. If you use the AI to guide you, balance ideas off, develop and run with a plot to see interest dynamics, that's just smart adoption and leveraging technology proactively. The argument AI is bad because it's AI, which is normally all the argument really boils down to, is terrible. AI is a tool, tools have a use (generally), and why not leverage good tools?

I have a multi-tool in my garbage, screwdrivers, plier, small hammer tool (actually several). I also have hammers, screwdrivers, drills, impact drivers, a drill press, and a pretty nice equipped workshop. Would using the multi-tools void what I can do with the other tools? Would using the other tools subtract value from the multi-tool? Maybe a better comparison would be building a jig, would that take away from my tools? If I need a jig, wouldn't that be a crutch to help me, vs having to free hand the same design multiple times?

My teenage daughters look physically ill when you bring up AI. “Ewww, not AI, I hate that so much, just do the work yourself!”, really? You can't do your math homework without a calculator and countless hours of confusion, why is the calculator not the same type of crutch? People hate AI for the same reason they hate people who can figure out a solution without excessive complaining, crying and hissy-fitting.

Comment Re:Identify the customers and issue refunds! (Score 1) 99

What large retailer doesn't have analytics on overdrive, being fed into AI-powered ingestion to track every consumer? Call a customer service line and hear the classic (paraphrased): “We record and track all customer interactions for improvements and training.”, then, if you ask for that recording, silence, rejection, excuses. 99%, and I think that's low, of companies know what you bought, when and how it related to your purchase history and power. We've heard the excuses and rational for years and decades, now use them... just and idea. I've said something similar to “Data collection only exists to abuse and misuse data.”, and yet again, proven right, why is that data not being used to help?

Comment You trusted Microsoft, and email? (Score 5, Insightful) 78

Email is not reliable, at best it's a fractured system of different protocols built on top of numerous other protocols that all have to work. Normally, email does work, and it's not an issue, but, just because something does normally work doesn't make it a reliable and trustable system. Take that, and introduce Microsoft, and what did anyone expect to happen?

As stated many, many times, why was their no physical alert? Even if the email system worked correctly, and by some miracle chance Microsoft didn't screw up, big, big ask, there should still be a physical alert. You have to assume your notification / alerting system is down, has malfunctioned or failed, and that's why we build redundancy.

To the engineers responsible, or not responsible since they were negligence (could be management), you need to build in N level redundancy. The email as a nice to have is great, but it's a nice to have extra, not a safe reliable and trusted notification system for emergencies. Before the email, there should have been a visual alert, audio-based alert, and complex / arena wide computer-based PID alert. All three, at least, should have triggered, and the email notification if one existed, then it can notify, but under no circumstance should email have been trusted in this way.

Comment Identify the customers and issue refunds! (Score 2) 99

Every major store or site has monitoring. It might be camera-based, likely AI based, log based, email based, and so on, and certainly was sold as a way to protect the companies, and, improve the user / consumer experience. If you have all of this information, use it, find the customers, correlate invoices, recipes, purchases, and cut a cheque. Return all tariff related expenses which were passed to the consumer, they know what they are, and were, have the records to prove it, so get to work.

I understand how much work this is, it's headache inducing, and approaches what sounds like an impossible task, but, companies effectively stole the money. Where's the money going to come from to fund this? The companies themselves, without pricing increases, and without fee's, adjustments, and tax kickbacks. The money was literally illegally collected, and illegally handled, so why can the refunds be effectively illegally used?

Comment Re: Stop dressing it up (Score 1) 57

It might not be 100%, but it's in the high 90's, and that's enough to sell it through. Sitting here I can't think of a single service or product Microsoft offers that could reach an aluminum foil standard. Forget bronze, what does Microsoft offer that could reach aluminum? Their office suite has been a joke for years. Their operating system is such a mess, they have to let Linux run on top of it, so people can do work. Azure is flaming dumpster fire, that's being kind, and I work in Azure a fair amount, it's horrific. Their productivity tools, have you tried PowerBI or Project? I can't think of a single productivity tool outside what I mentioned worth running. Development tools? All terrible, across the board.

If you want to get into languages / frameworks, .Net is a joke, C# / Entity is a disgrace. I'll give some marks to Typescript because it's actually decent, maybe up to a tarnish aluminum foil standard.

Honestly, what product or service can you look at and honestly say: "Bronze standards or higher", I can't think of a single one.

Slashdot Top Deals

Save energy: Drive a smaller shell.

Working...